Fan blade terahertz imaging detection method based on MIMO array

Abstract

The present application relates to the technical field of terahertz radar imaging and nondestructive testing, and particularly relates to a fan blade terahertz imaging detection method based on a MIMO array, steps of which are as follows: a terahertz MIMO array is carried by a motion detection platform to perform motion scanning around a fan blade to obtain terahertz echo data; the terahertz echo data is subjected to equidistance slice imaging processing to obtain a plurality of terahertz slice images; the plurality of terahertz slice images are subjected to maximum non-coherent fusion to obtain a terahertz fusion image; and the terahertz fusion image is input into a pre-trained damage detection network to obtain a blade damage detection result. The present application improves the terahertz imaging lateral resolution without significantly increasing the physical array size, and can reduce the artificial detection risk and detection implementation difficulty, and has a higher imaging processing efficiency and a lower missed detection risk, and has stronger objectivity and consistency compared with artificial rule interpretation.

Description

Technical Field

[0001] This invention relates to the field of terahertz radar imaging and non-destructive testing technology, and more specifically, to a terahertz imaging detection method for wind turbine blades based on a MIMO array. Background Technology

[0002] Wind turbine blades are key components in wind power generation equipment that bear aerodynamic loads, gravity loads, centrifugal loads, and complex environmental loads. During long-term service, blades are prone to defects such as cracks, delamination, debonding, internal voids, inclusions, water accumulation, lightning strike damage, and impact damage. If these defects are not detected in time, they may lead to a decrease in the aerodynamic performance of the blades, a reduction in structural strength, or even a breakage accident. Therefore, it is of great significance to inspect wind turbine blades.

[0003] Existing wind turbine blade inspection technologies mainly rely on non-destructive testing, including manual visual inspection, high-definition camera inspection, infrared thermal imaging, ultrasonic testing, acoustic emission testing, microwave testing, and laser scanning testing. Among these, manual visual inspection and optical inspection methods depend on lighting conditions and surface visibility. Their reliability decreases under conditions such as rain, fog, smoke, low light, strong reflection, and blade surface contamination, and they are difficult to identify internal blade defects. Ultrasonic testing has high requirements for coupling conditions and contact state, making it difficult to implement in large-sized curved blades and high-altitude operation scenarios. Infrared thermal imaging is easily affected by ambient temperature, wind speed, solar radiation, and material thermal diffusion characteristics, and its ability to identify deep or small-scale defects is limited.

[0004] Terahertz waves, located between microwaves and infrared, possess a certain ability to penetrate non-metallic materials and have high spatial resolution potential, making them suitable for internal structure inspection of composite wind turbine blades. Traditional terahertz detection methods often employ single-site scanning or small-aperture imaging, which have limited detection range, insufficient two-dimensional resolution, and low scanning efficiency, making it difficult to meet the inspection needs of large-sized wind turbine blades. For high-altitude areas on the outer surface of blades, narrow channels inside blades, and dangerous areas that are difficult for personnel to access, fixed or handheld terahertz detection equipment also faces difficulties in deployment and insufficient detection efficiency.

[0005] MIMO (Multiple Input Multiple Output) arrays can form a virtual array by combining multiple transmitting elements and multiple receiving elements, thereby expanding the equivalent aperture under the condition of limited physical array size. If further combined with motion scanning, the array aperture and the scanning aperture can be combined to form a two-dimensional large aperture or a three-dimensional synthetic aperture, which is beneficial to improving the lateral resolution and coverage of terahertz imaging of wind turbine blades.

[0006] Therefore, there is an urgent need to provide a terahertz imaging detection method for wind turbine blades that can adapt to complex external environments and narrow detection spaces, while taking into account high-resolution imaging, high scanning efficiency and damage identification capabilities. Summary of the Invention

[0007] The purpose of this invention is to provide a terahertz imaging detection method for wind turbine blades based on MIMO arrays, so as to solve the technical problems pointed out in the background art.

[0008] This invention is achieved through the following technical solution: a terahertz imaging detection method for wind turbine blades based on MIMO arrays, comprising the following steps: A terahertz MIMO array is configured, the terahertz MIMO array comprising multiple transmit elements and receive elements; The motion detection platform, equipped with the terahertz MIMO array, performs motion scanning around the wind turbine blades to acquire terahertz echo data; The terahertz echo data is subjected to equidistant slice imaging processing to obtain several terahertz slice images; A maximum value incoherent fusion is performed on several of the terahertz slice images to obtain a terahertz fused image; The terahertz fused image is input into a pre-trained damage detection network to obtain the blade damage detection results.

[0009] According to a preferred embodiment, after acquiring the terahertz echo data, the method further includes: Based on the pose information acquired in real time during motion scanning, motion error compensation and spatial coordinate unification processing are performed on the terahertz echo data.

[0010] According to a preferred embodiment, after acquiring the terahertz echo data, the method further includes: The terahertz echo data is preprocessed, including at least one of the following: DC component removal, time synchronization, range-direction matched filtering, channel amplitude and phase compensation, abnormal pulse rejection, noise suppression, and scan position resampling.

[0011] According to a preferred embodiment, the terahertz echo data is subjected to equidistant slice imaging processing, specifically including: The preprocessed terahertz echo data is transformed to the wavenumber domain to obtain non-uniform wavenumbers, including range wavenumbers obtained from the operating frequency and transverse wavenumbers obtained from the aperture coordinates. Non-uniform wavenumber sampling is mapped to a regular wavenumber grid through Stolt interpolation, and then inverse Fourier transform is performed to obtain the imaging volume data; The imaging data is sliced ​​according to the equidistant depth position to obtain several terahertz slice images, wherein each terahertz slice image corresponds to the scattering intensity distribution on the wind turbine blade at different depths or distances.

[0012] According to a preferred embodiment, maximum incoherent fusion is performed on a plurality of the terahertz slice images, specifically including: For each spatial pixel location, the maximum amplitude or power value of each spatial pixel location is taken from several terahertz slice images, and the values ​​of each spatial pixel location are stitched together to obtain a terahertz fused image.

[0013] According to a preferred embodiment, after obtaining the terahertz fused image, the method further includes: The enhancement processing of the terahertz fused image includes at least one of contrast enhancement, background equalization, artifact suppression, size normalization, and image cropping.

[0014] According to a preferred embodiment, the transmitting array element and the receiving array element are equivalent to a virtual array, and the equivalent array element spacing of the virtual array is no greater than half of the wavelength corresponding to the working center frequency.

[0015] According to a preferred embodiment, the terahertz MIMO array operates in strip mode, and the transmitted terahertz detection signal is a stepped-frequency continuous wave signal, a frequency-modulated continuous wave signal, a pulse signal, or any combination thereof, with an operating frequency range of 0.1THz to 10THz.

[0016] According to a preferred embodiment, the motion detection platform is a drone, a wall-climbing robot, a rail-mounted scanning platform, a rope traction platform, a flexible attachment platform, or a mobile carrier that can enter the interior of a wind turbine blade.

[0017] According to a preferred embodiment, the motion scanning includes longitudinal scanning along the length of the wind turbine blade, transverse scanning along the chord direction of the wind turbine blade, serpentine reciprocating scanning, partitioned splicing scanning, or surface scanning planned according to the curved surface morphology of the wind turbine blade.

[0018] The technical solution of the terahertz imaging detection method for wind turbine blades based on MIMO arrays provided by this invention has at least the following advantages and beneficial effects: This invention forms a two-dimensional large aperture by combining a MIMO array with motion scanning, improving the lateral resolution of terahertz imaging without significantly increasing the physical array size, thus addressing the problem of insufficient resolution in traditional solid-aperture terahertz imaging for large-size wind turbine blade detection; by mounting the terahertz MIMO array on a motion detection platform, it can detect high-altitude areas on the outer surface of the blade, narrow areas inside the blade, and high-risk areas difficult for personnel to reach, thereby reducing the risk and difficulty of manual detection; the use of a fast distance migration algorithm for equidistant slice imaging can adapt to multi-dimensional echo data formed by the combination of array aperture and motion scanning aperture, thereby improving imaging processing efficiency and facilitating rapid on-site detection; by performing maximum noncoherent fusion of multiple equidistant terahertz slice images, it can enhance the abnormal response of wind turbine blade damage locations and reduce the risk of missed detection caused by single slice selection; by combining the high-resolution terahertz imaging results with a damage detection network, it can output damage type, location, and confidence level, which has stronger objectivity and consistency compared to manual rule-based interpretation. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall process of the terahertz imaging detection method for wind turbine blades based on MIMO array provided in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of a virtual array provided in Embodiment 1 of the present invention; Figure 3 This is a schematic diagram of motion scanning provided in Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of maximum noncoherent fusion provided in Embodiment 1 of the present invention; Figure 5 This is a schematic diagram of the damage detection network provided in Embodiment 1 of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0021] Example 1 This invention provides a terahertz imaging detection method for wind turbine blades based on a MIMO array. Figure 1 This is a schematic diagram of the overall process of the terahertz imaging detection method for wind turbine blades, as shown below. Figure 1As shown, the terahertz imaging detection method for wind turbine blades includes the following steps: Step S1: Configure the terahertz MIMO array; In this embodiment, the terahertz MIMO array includes multiple transmitting elements and receiving elements, for example, configured as follows: Each launch element and Each receiving array element is excited by time-division multiplexing, frequency-division multiplexing, code-division multiplexing, or orthogonal waveform multiplexing, and the echo signal is synchronously received by multiple receiving array elements to form a signal. There are 1 equivalent transmit / receive channel, where time division multiplexing is used for sequential excitation, and frequency division / code division / orthogonal waveform multiplexing is used for simultaneous excitation.

[0022] Furthermore, the transmitting and receiving array elements are equivalent to a virtual array, such as... Figure 2 As shown, the equivalent element spacing of the virtual array Not greater than the wavelength corresponding to the working center frequency Half of; for scenarios requiring higher anti-aliasing margin, the equivalent element spacing It can also be set to 0.4 to 0.5 times the wavelength.

[0023] In some preferred embodiments, the terahertz MIMO array operates in strip mode, and the transmitted terahertz detection signal is a stepped-frequency continuous wave signal, a frequency-modulated continuous wave signal, a pulse signal, or any combination thereof, with an operating frequency range of 0.1 THz to 10 THz; taking the frequency-modulated continuous wave signal as an example, the terahertz detection signal has a bandwidth within a single sweep cycle. The range resolution can be approximated as: ,in Since the speed of light is so high, by increasing the bandwidth, the resolution in the thickness direction and the defect depth direction of the blade can be improved.

[0024] Step S2: Set the motion trajectory and speed of the motion detection platform; The terahertz MIMO array is mounted on a motion detection platform. For the detection of the outer surface of the blade, the motion detection platform can be a drone, a wall-climbing robot, a rail-mounted scanning platform, or a flexible attachment platform. For the detection of the internal channels of the blade, the motion detection platform can be a small tracked vehicle, a wheeled robot, or a rope-traction platform, or any other mobile carrier that can enter the interior of the wind turbine blade. No specific restrictions are imposed here.

[0025] Furthermore, a scanning coordinate system is established based on the area to be inspected on the blade: defined along the blade length direction as... Direction, defined as the chord-direction The direction, perpendicular to the local surface of the blade, is defined as... Direction. The motion detection platform moves along a preset strip path, keeping the distance between the array and the blade surface within a set range at each sampling position, and keeping the array normal pointing towards the local surface of the blade.

[0026] The motion speed and adjacent sampling interval are determined based on the azimuth resolution and scanning sampling conditions; preferably, the adjacent sampling interval is no greater than half of the working wavelength or no greater than half of the desired lateral resolution, so as to ensure sufficient spatial sampling of the motion synthetic aperture; the platform motion speed can be determined by multiplying the adjacent sampling interval by the echo acquisition frame rate.

[0027] It should be noted that this invention forms a two-dimensional large aperture by combining a MIMO array with motion scanning, which improves the lateral resolution of terahertz imaging without significantly increasing the physical array size, thus addressing the problem of insufficient resolution in traditional solid aperture terahertz imaging for the inspection of large-size wind turbine blades. By mounting the terahertz MIMO array on a motion inspection platform, it is possible to inspect high-altitude areas on the outer surface of the blade, narrow areas inside the blade, and high-risk areas that are difficult for personnel to reach, thereby reducing the risks and difficulties of manual inspection.

[0028] Step S3: Terahertz echo data acquisition and preprocessing; The motion detection platform, equipped with a terahertz MIMO array, performs motion scanning around the wind turbine blades. Specifically, the motion scanning includes longitudinal scanning along the length of the wind turbine blades, transverse scanning along the chord direction of the wind turbine blades, serpentine reciprocating scanning, partitioned splicing scanning, or surface scanning planned according to the curved surface morphology of the wind turbine blades.

[0029] In some preferred embodiments, such as Figure 3 As shown, the motion detection platform, equipped with a terahertz MIMO array, moves at close range along the blade surface. The moving path is divided into multiple strips along the blade length direction. After each strip is scanned, the motion detection platform, equipped with a terahertz MIMO array, moves one strip interval along the chord direction and then performs the next strip scan in the opposite direction, thus forming a serpentine coverage path. The imaging results of different strips are then stitched together to obtain a complete blade image.

[0030] At each scanning position, the terahertz MIMO array is controlled to transmit terahertz detection signals and receive reflected or scattered echoes from the wind turbine blade surface, skin interior, adhesive layer, sandwich structure, and damaged areas. Further, after mixing, sampling, and digitization, frequency domain or time domain echo data corresponding to each transceiver channel can be obtained. Terahertz echo data can be represented as... Its dimensions include the transmit channel dimension, receive channel dimension, frequency sampling dimension, and scan position dimension; for the case of using frequency-modulated continuous wave system, the frequency sampling dimension corresponds to multiple fast time sampling points or frequency points; for the case of using stepped frequency continuous wave system, the frequency sampling dimension corresponds to multiple discrete frequency points.

[0031] While acquiring echo data, the motion detection platform can record timestamps, poses, distances from the array to the blade surface, and scan strip numbers. This information is used for subsequent motion error compensation, surface coordinate correction, and defect location inversion. In some preferred embodiments, after acquiring terahertz echo data, motion error compensation and spatial coordinate unification processing can be further performed on the terahertz echo data based on the pose information acquired in real time during motion scanning. In addition, the preprocessing of the terahertz echo data also includes at least one of the following: DC component removal, time synchronization, range-matched filtering, channel amplitude and phase compensation, abnormal pulse rejection, noise suppression, and scan position resampling.

[0032] Secondly, the preprocessed terahertz echo data is mapped to equivalent two-dimensional aperture coordinates according to the virtual array element positions and scanning positions; for each virtual array element and each scanning position, the corresponding equivalent sampling point coordinates are determined by the array geometric position, platform pose and scanning path.

[0033] Next, a near-field imaging model is established, assuming the coordinates of a point within the imaging region are... That point to the The propagation distance of each equivalent sampling point is The echo phase can be approximated by the propagation distance. The decision is made by performing phase compensation and coherent superposition on echoes of different frequencies, different virtual array elements, and different scanning positions to form an imaging result.

[0034] Step S4: Isochronous slice imaging processing; To reduce the computational complexity of direct back projection imaging, this embodiment employs a fast distance migration algorithm to perform equidistant slice imaging processing on the terahertz echo data, thereby obtaining several terahertz slice images.

[0035] In some preferred embodiments, the specific components include: The preprocessed terahertz echo data is transformed to the wavenumber domain to obtain non-uniform wavenumbers, including range wavenumbers obtained from the operating frequency and lateral wavenumbers obtained from the aperture coordinates. Stolt interpolation is used to map the non-uniform wavenumber samples to a regular wavenumber grid, followed by inverse Fourier transform to obtain the imaging volume data. The imaging volume data is sliced ​​according to equidistant depth positions to obtain several terahertz slice images. Each terahertz slice image corresponds to the scattering intensity distribution at different depths or range gates on the wind turbine blade. It is worth noting that Stolt interpolation is a key technology for synthetic aperture radar imaging, mainly used in wavenumber domain imaging algorithms. It achieves range-frequency axis mapping through variable substitution, thereby enabling focusing on non-reference range targets and solving the problems of range migration and residual phase compression.

[0036] It should be noted that for curved blades, tangent plane coordinates can be pre-established in a local area to complete slice imaging, and then global stitching can be performed based on the platform pose and the blade surface model. This will not be elaborated further here. This embodiment uses a fast distance migration algorithm for equidistant slice imaging, which can adapt to multi-dimensional echo data formed by the combined array aperture and moving scanning aperture, thereby improving imaging processing efficiency and facilitating rapid on-site inspection.

[0037] Step S5: Perform maximum incoherent fusion; In this embodiment, maximum incoherent fusion is performed on several terahertz slice images to obtain a terahertz fused image; specifically including: Figure 4 As shown, for each spatial pixel location, the maximum amplitude or power value of each spatial pixel location is taken from several terahertz slice images, and the values ​​of each spatial pixel location are stitched together to obtain a terahertz fused image; let the th... A terahertz slice image at pixel location The amplitude at the point is The resulting terahertz fused image at that location will have multiple values. The maximum value in is represented as .

[0038] It should be noted that using maximum incoherent fusion can preserve the most significant anomalous scattering responses in terahertz slice images at each depth, enhancing the image features of crack tips, delamination boundaries, adhesive debonding edges, internal void boundaries, and water accumulation areas. At the same time, since the fusion process does not depend on the phase consistency between slices, it can reduce the impact of motion errors, material inhomogeneities, and surface attitude errors on the fusion results.

[0039] Furthermore, after obtaining the terahertz fused image, further image enhancement processing can be performed. Image enhancement processing includes at least one of contrast enhancement, background equalization, artifact suppression, size normalization, and image cropping. The enhanced terahertz fused image is used as input for subsequent damage detection networks.

[0040] It should be noted that this embodiment can enhance the abnormal response of the wind turbine blade damage location by performing maximum value incoherent fusion on multiple equidistant terahertz slice images, thereby reducing the risk of missed detection caused by single slice selection.

[0041] Step S6: Damage detection; In this embodiment, as Figure 5 As shown, a damage detection network is used for detection. The terahertz fusion image after image enhancement is input into the pre-trained damage detection network, which outputs the blade damage detection results, including each suspected damage category, location box and confidence score.

[0042] Specifically, the damage detection network can be the YOLO damage detection network, which includes a backbone feature extraction network, a neck multi-scale feature fusion network, and a detection head. The detection head outputs the suspected damage category, location box, and confidence score at different scales. When training the damage detection network, a terahertz damage image dataset of wind turbine blades can be constructed. This dataset includes samples of cracks, delamination, debonding, impact damage, internal voids, water accumulation, foreign object inclusions, lightning strike damage, and normal areas. The sample sources can be real wind turbine blade inspection images, artificial defect specimen inspection images, simulation images, and data-enhanced images processed by rotation, scaling, cropping, brightness perturbation, and noise overlay.

[0043] Furthermore, after obtaining the blade damage detection results, the image coordinates can be converted into the physical coordinates of the blade by combining the scanning coordinate system and the pose information of the motion platform, thereby obtaining the actual location and range of the damage on the wind turbine blade and generating a corresponding detection report.

[0044] In summary, this embodiment combines terahertz high-resolution imaging results with a damage detection network to output damage type, location, and confidence level, which has stronger objectivity and consistency compared to manual rule-based interpretation.

[0045] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A terahertz imaging detection method for wind turbine blades based on MIMO arrays, characterized in that, Includes the following steps: A terahertz MIMO array is configured, the terahertz MIMO array comprising multiple transmit elements and receive elements; The motion detection platform, equipped with the terahertz MIMO array, performs motion scanning around the wind turbine blades to acquire terahertz echo data; The terahertz echo data is subjected to equidistant slice imaging processing to obtain several terahertz slice images; A maximum value incoherent fusion is performed on several of the terahertz slice images to obtain a terahertz fused image; The terahertz fused image is input into a pre-trained damage detection network to obtain the blade damage detection results.

2. The terahertz imaging detection method for wind turbine blades based on MIMO array as described in claim 1, characterized in that, After acquiring the terahertz echo data, the method further includes: Based on the pose information acquired in real time during motion scanning, motion error compensation and spatial coordinate unification processing are performed on the terahertz echo data.

3. The terahertz imaging detection method for wind turbine blades based on MIMO array as described in claim 2, characterized in that, After acquiring the terahertz echo data, the method further includes: The terahertz echo data is preprocessed, including at least one of the following: DC component removal, time synchronization, range-direction matched filtering, channel amplitude and phase compensation, abnormal pulse rejection, noise suppression, and scan position resampling.

4. The terahertz imaging detection method for wind turbine blades based on MIMO array as described in claim 3, characterized in that, The terahertz echo data is subjected to equidistant slice imaging processing, specifically including: The preprocessed terahertz echo data is transformed to the wavenumber domain to obtain non-uniform wavenumbers, including range wavenumbers obtained from the operating frequency and transverse wavenumbers obtained from the aperture coordinates. Non-uniform wavenumber sampling is mapped to a regular wavenumber grid through Stolt interpolation, and then inverse Fourier transform is performed to obtain the imaging volume data; The imaging data is sliced ​​according to the equidistant depth position to obtain several terahertz slice images, wherein each terahertz slice image corresponds to the scattering intensity distribution on the wind turbine blade at different depths or distances.

5. The terahertz imaging detection method for wind turbine blades based on MIMO array as described in claim 4, characterized in that, Performing maximum incoherent fusion on several terahertz slice images specifically includes: For each spatial pixel location, the maximum amplitude or power value of each spatial pixel location is taken from several terahertz slice images, and the values ​​of each spatial pixel location are stitched together to obtain a terahertz fused image.

6. The terahertz imaging detection method for wind turbine blades based on MIMO array as described in claim 5, characterized in that, After obtaining the terahertz fused image, the method further includes: The enhancement processing of the terahertz fused image includes at least one of contrast enhancement, background equalization, artifact suppression, size normalization, and image cropping.

7. The terahertz imaging detection method for wind turbine blades based on MIMO array as described in claim 1, characterized in that, The transmitting and receiving array elements are equivalent to a virtual array, and the equivalent array element spacing of the virtual array is no greater than half of the wavelength corresponding to the working center frequency.

8. The terahertz imaging detection method for wind turbine blades based on MIMO array as described in claim 1, characterized in that, The terahertz MIMO array operates in strip mode, and the transmitted terahertz detection signal is a stepped-frequency continuous wave signal, a frequency-modulated continuous wave signal, a pulse signal, or any combination thereof, with an operating frequency range of 0.1THz to 10THz.

9. The terahertz imaging detection method for wind turbine blades based on MIMO array as described in claim 1, characterized in that, The motion detection platform can be a drone, a wall-climbing robot, a rail-guided scanning platform, a rope traction platform, a flexible attachment platform, or a mobile carrier that can enter the interior of the wind turbine blades.

10. The terahertz imaging detection method for wind turbine blades based on MIMO array as described in claim 1, characterized in that, The motion scanning includes longitudinal scanning along the length of the wind turbine blade, transverse scanning along the chord direction of the wind turbine blade, serpentine reciprocating scanning, partitioned splicing scanning, or surface scanning planned according to the curved surface morphology of the wind turbine blade.

Images